A new survey finds 68% of US adults use an AI agent weekly, but few trust one with a credit card, and many are unsure who pays if it goes wrong.
Most Americans now use AI helpers, but they are not ready to let them spend money. A survey of 2,026 US adults by fraud prevention firm SEON found that 68% use an AI agent for a personal task at least weekly. Only 15% would share financial details with one, and 32% would share nothing at all.
An AI agent is a tool that does tasks for you, not just answers questions. Half of respondents said they would feel uneasy letting one use their credit card, even for a purchase they asked for. Unease was highest among Boomers at 65% and lowest among Millennials at 34%.
People are more willing to share other data. Location tops the list at 38%. Purchase history and calendar access are both at 34%. Younger people were more open to sharing financial details, but still a minority: 23% of Millennials and 22% of Gen Z, against 4% of Boomers.
If an agent completes a fraudulent purchase, 65% said the AI company should be responsible. Only 14% blamed the shopper, 11% the merchant and 10% the bank or card issuer. That is what people believe, not a ruling on who actually pays. The material does not say how banks or courts would treat these cases, so the rules should be treated as unsettled.
Fear of being fooled is part of the picture. 43% doubt they could tell whether an agent's request is genuine, rising to 61% of Boomers. And 39% do not trust agents to flag risky situations. The top fears, at 20% each, were a scammer hijacking the agent or their accounts, and the agent sharing data without consent.
Anyone using an assistant that can shop or pay is in the group to watch. So are banks, payment companies and online merchants, which will need to tell a real agent acting for a customer from a criminal using one. SEON points to tools such as Meta's Muse and OpenAI's Dots as moving agents closer to acting on their own.
One caution on the source. SEON sells fraud prevention software, so it has an interest in this topic. The press release gives no more detail on how the survey was run.
The alert box above lists the steps. The short version: give an agent only what a task needs, and pick one that offers real help from a person and a way to undo mistakes. Those two features matter most to shoppers. Human support was cited by 50% as the top trust builder, and the ability to undo errors by 42%.
Expect more agents that can act without asking each time, and more pressure on companies to show who is liable when something goes wrong. For now, 23% of those surveyed said nothing would win their confidence. If you are in that camp, keeping payments out of agents' hands is a reasonable choice.
Before an AI agent touches your card, make sure you can reach a human and undo a mistake, because who pays for agent fraud is still unclear.
Fake PayPal emails and texts, some posing as a security report, try to get you to tap a link and type your login. Checking your account directly avoids the trap.
Read more: Got a fake PayPal message? Skip its link and check your account yourselfSecurity firm Flashpoint says criminals are trading in refund tricks, gift card fraud and hacked online stores ahead of the holidays. Shop owners should check their refund steps.
Read more: Scammers are targeting store refunds and gift cards before the holidaysA user says Anthropic's Claude Code put his email in web requests, then scam messages arrived. The link is unproven, but it shows what AI tools can share.
Read more: User says Claude Code sent his email address to websites
